Laura Savage started off the discussion by looking back at the ten years that have passed since the beginning of RISE. She noted that some questions in education have persisted throughout this time: for example, how do you successfully go to scale? How do you create the political environment in which actors will support effective approaches to improve learning?

Nompumelelo Mohohlwane said that she has encountered a number of similarly yet-unanswered questions in her position overseeing education research across South Africa. Beyond questions on topics such as literacy, violence, parental involvement, and nutrition that are part of her department’s research agenda, persistent questions that her department continues to consider include: at what age should schools switch language of instruction to English? How long-lasting are the effects of early interventions? And how much do young South African children actually know? Although South African students in the early grades have participated in many national and international assessments, said Mohohlwane, more details about what children do know are needed.

As to why these questions haven’t yet been answered, Mohohlwane said some of the reasons are ideological, some are around what evidence exists and is available, and some are around political economy. Governments must decide where to focus their energy, and they may face competing priorities that are both important, such as unemployment and education.

Laura Savage pointed out that some of the conversations about these ongoing questions involve the issue of who should be answering them. Economists, anthropologists, and experts in public administration can approach old problems from new angles.

Rachel Hinton agreed that RISE’s use of an interdisciplinary group of researchers and collaborators has had great effect and has shifted thinking about approaches to education research. She shared a graph exploring how different research approaches can and have helped address gaps in education research and offered important insights.

Ten years ago, Hinton said, studies tended to be qualitative and on the micro scale, not looking at education through a systems lens. Overall, the focus was more on access than on learning.

However, RISE has been a part of a shift in approaches toward the macro scale and quantitative research, as well as a shift toward focusing on learning. In recent years, the Smart Buys report and the work of the Global Evidence in Education Advisory Panel show this development in research approaches and the great increase in available evidence.

RISE has also changed the narrative on evidence uptake, said Hinton, and that’s partly due to the different voices within RISE. For example, the RISE framework was influential for the World Development Report 2018, and the Indian government’s 2020 National Education Policy was very informed by research from members of the RISE India team such as Abhijeet Singh and Karthik Muralidharan.

What needs to happen next in terms of research approaches? Hinton said there’s a need for more mixed methods research and for greater empowerment of local voices and researchers from the Global South.

The motivating question of RISE, said Lant Pritchett, was: why do some education systems, as a result of the routine operation of the system, endogenously produce excellent learning outcomes while other systems don’t? That’s not a question about how to scale up individual components or programmes, but a question about systems.

Pritchett raised the point that Vietnam has famously achieved high learning outcomes at lower costs per pupil. How has Vietnam done this? Pritchett said RISE has made progress in answering the question in terms of the characteristics of functional systems, although it hasn’t focused as much on how specific results have been achieved—as there are always numerous ways to reach the same educational goal.

This is where implementation science comes in. Pritchett explained that there are two different interpretations of the term “implementation science”:

• “How do we get governments to implement what we think they should be implementing to achieve desired outcomes?”
• “What is the science of systems that work such that they implement well and achieve the outcomes that we want?”

Pritchett felt that there’s a need for implementation science in the second sense. “We don’t actually have a science of ‘how does progress happen?’” he said.

Pritchett and Laura Savage agreed that both systems thinking and implementation science shouldn’t be taken to mean explorations of how external actors can mould a system into their desired shape. Both systems thinking and implementation science are about the characteristics within the system, such as the purpose and commitment within the system that’s needed to drive change.

This is why, Pritchett said, the challenges faced by the WWHGE are so much greater than those faced by RISE. RISE’s task was to produce research, and it did that well. But the WWHGE needs to know how to produce commitment, and commitment has to be created locally.

Nompumelelo Mohohlwane said that there is commitment to learning in South Africa. The point to address may be different ideas about the best way to achieve learning. The question she identified is: how can those ideas be shifted?

Mohohlwane gave two examples. One is a design failure—an intervention that was designed to give parents and teachers information that would lead them to make different choices. But principals didn’t welcome the information that their school was doing less well than neighbouring schools and didn’t want to continue taking part. Was this a sign of a lack of commitment to learning? No, said Mohohlwane, the issue was that this lever didn’t build credibility for the school principal.

A more successful intervention was one in which training was given within a number of constraints, including budget and time cycle constraints. These constraints also weren’t due to a lack of commitment to learning, Mohohlwane explained. Importantly, what made the intervention successful was that it was implemented flexibly within the constraints of the system.

Laura Savage asked: what’s the key to working within constraints at the design and testing stage? How do you know which constraints are movable?

The key, said Mohohlwane, was that the programme’s implementers were within the state, with a strong understanding of the players involved. There was trust among stakeholders, and Mohohlwane’s team listened to local advisors’ opinions about what they thought would work. Implementing as true partners was important.

Lant Pritchett noted that implementation isn’t established yet as a discipline, and that’s a problem. What counts as a question or an answer in implementation science hasn’t yet been defined. And because answers in implementation may not involve causal identification, it’s difficult to frame implementation questions in a way that other disciplines will recognise.

Rachel Hinton said that implementation science will need to examine education systems as whole entities, including bureaucratic culture, incentives for change, social constraints, and equity. When interventions haven’t had an impact on learning, implementation science can come in and examine what led to the lack of impact and what could be shifted.

Many thanks to all who took part in this great conversation and the RISE Annual Conference 2023 as a whole. To see more from this session, including the Q & A section, stop by our YouTube channel.

Although, sadly, the RISE Annual Conference 2023 was the last RISE conference, RISE’s extensive library of research and tools will remain available. We’re looking forward to the endeavours of the What Works Hub for Global Education, which is now in its very early phases. Watch out for news from the University of Oxford’s Blavatnik School of Government to learn about the What Works Hub for Global Education’s first ventures—including plans for a conference next year in place of RISE.